Commit Graph

8 Commits

Author SHA1 Message Date
b33829ded5 Add skeleton loading, self-host assets, fix LLM empty response
Perceived startup speed:
- index.html: animated shimmer skeleton (header + toolbar + 8 voice cards)
  visible immediately; fades out when loader.js finishes
- WaveSurfer (57KB) and MDI icon font (394KB woff2) now served from
  static/vendor/ — removes 3 render-blocking external requests from <head>
- Flag-icons CSS loaded async (rel=preload onload trick) — non-blocking

loader.js:
- Fade out skeleton + remove from DOM (300ms transition)
- Reveal page-sections after JS finishes loading

Conversation playground:
- Qwen3 thinking mode fix: fall back to delta.reasoning_content when
  delta.content is empty so think-only LLM turns produce visible output
- Better error message with /no-think hint when LLM returns empty

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 17:28:42 +02:00
a5efac5da5 Speed up startup: parallel JS loading + versioned asset caching
loader.js:
- Fetch /api/version first; append ?v=<version> to every script URL
- Load scripts with async=false (fetch in parallel, execute in DOM order)
- 4 sequential batches instead of 17 sequential awaits:
    utils → settings → [9 feature modules] → init → [4 post-init] → nav
- 17 round-trips → 6 round-trips; 9 and 4 files now download in parallel
- Section HTML keeps ?v=<timestamp> (no-store, always fresh)

server.py:
- Add static_cache_headers middleware:
    /static/js/*?v=...  → Cache-Control: max-age=31536000, immutable
    /static/sections/*  → Cache-Control: no-store
    /static/* (no v=)   → Cache-Control: max-age=3600

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 17:11:55 +02:00
74e5181b7e Fix conversation input bar hidden when mic is blocked
- Move mic-blocked warning into conv-chat-window (prepend) so it scrolls
  with the chat instead of pushing the input bar off-screen
- Remove min-height:400px from conv-chat-window; give conv-chat-panel a
  viewport-relative height so the input bar is always visible at the bottom
- loader.js: append ?v=<timestamp> to section fetches to bust browser cache
  (was serving stale s-conversation.html after updates)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 14:20:52 +02:00
c7a1e35539 Security audit, modular refactor, and container-name field
Security fixes:
- Block /proc /sys /dev /run /boot in /api/browse-dirs (path traversal)
- Verify yt-dlp output stays inside TEMP_DIR before registration
- Remove Access-Control-Allow-Origin: * from /api/proxy-audio
- TTL-based temp file registry (default 2h) to prevent disk fill

Performance:
- Cache settings + routing rules in memory (mtime-checked); eliminates
  per-request disk reads on every TTS call

UI:
- Add container name (optional) field to Docker stack TTS/STT engine
  cards (Qwen3 Voice Clone, Voice Design, Custom Voice, Streaming,
  NVIDIA Magpie, Parakeet) — enables Stop/Start/Restart buttons on
  all engine cards, matching the existing Other Local TTS/STT cards

Refactor — backend:
- server.py: 5560 lines → 43-line entry point
- core/ package: constants, registry, validation, docker_client,
  config, routing, audio, voice, presets, tts_helpers
- routes/ package: admin, settings, library, stt, sources, docker,
  tts, conversation (FastAPI APIRouter modules)
- Dockerfile + docker-compose.yml updated to include core/ and routes/

Refactor — frontend:
- static/app.js: 8744 lines → 16 modules in static/js/
  utils, voice-inspector, voice-sources, integrations, routing,
  settings, voice-clone, voice-library, tts-preview, benchmark,
  stt, init, engines, ai-backends, generation, conversation
- static/loader.js updated to load modules sequentially

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-29 12:13:07 +02:00
6a78be7a28 Add Conversation Playground + XTTS v2 backend + VibeVoice voice fix
Conversation Playground (new section):
- WhatsApp-style chat UI with user/assistant speech bubbles
- Click-to-record mic button using MediaRecorder API
- STT → LLM streaming → TTS pipeline via SSE (POST /api/conversation/turn)
- LLM tokens stream into assistant bubble in real time
- Audio auto-plays when TTS synthesises the reply
- Right-side stats panel: STT / LLM TTFT / LLM total / TTS / Total with bar chart
- Turn history list with per-turn total time and pass/fail indicator
- Configurable: STT backend, LLM URL + model, TTS backend + voice, system prompt
- Conversation history maintained across turns (last 20 messages sent to LLM)
- GET /api/conversation/llm-models proxies model list from any OpenAI-compatible LLM

XTTS v2 backend:
- Registers xtts as a first-class TTS backend (xtts_url setting, display name,
  capabilities, health/voice discovery, OpenAI-compatible generation)
- Added XTTS URL field to Settings → Connections
- Use-as-TTS button now saves to xtts_url (not tts_url)
- Batch benchmark backend select now refreshes alongside perf/preview selectors

VibeVoice fix:
- Added /voices to _TTS_VOICE_ENDPOINTS so VibeVoice voices are discovered

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-27 23:27:57 +02:00
dc47aa0431 Add Benchmark page, MCP snippets, Captures settings, and history tracking
- New s-performance section: dedicated Benchmark nav entry with run form,
  per-session results table, RTF trend badge (faster/slower/stable), SVG
  sparkline chart, and a History card backed by localStorage (last 50 sessions)
- Performance tab removed from Try It Out; element IDs unchanged so JS works
- renderIntegrationSnippets: adds Python MCP server + Claude Code .mcp.json
  config snippets to the Connect Apps page (integration-card-wide styling)
- Save handler: persists all Captures settings fields (stt_language,
  stt_preferred_backend, auto_refine, refine_model, refine_* toggles,
  captures_default_voice) alongside existing settings
- CSS: integration-card-wide accent border, benchmark history rows, trend
  badges, sparkline wrapper, bench-history-toolbar

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-27 09:57:55 +02:00
28f5ec2e25 Redesign inspector pane, add searchable pickers, and style AI Backends tab
Inspector:
- Skinny 2-row header: 72px avatar + name/ID row / subtitle row / note row
- Searchable flag picker (dblclick flag icon) — filtered by voice language, falls back to ALL_FLAGS
- Searchable language picker (dblclick lang code) — shows full language names
- Tag reuse: entered tags persist to localStorage as datalist suggestions
- Compact active toggle (32×18px), slim save button (12px/4px padding)
- Show/hide eye toggle and "✓ Key saved" badge on API key fields

AI Backends (s-llms.html + style.css):
- Full CSS design: pill tabs with active accent, animated section transitions
- Service cards: icon bubbles, tier badges (Free/Demo/Paid), stat chips, endpoint rows, model tags
- Highlighted recommended card with accent border
- Dark code blocks for local service snippets with copy feedback
- Show/hide password toggle and auto-appearing "✓ Key saved" badge per card

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-26 11:23:27 +02:00
631ca16290 Studio Notebook redesign: warm cream theme, sidebar, multi-file static structure
- New light UI: fixed 220px sidebar, single scrolling page, 8 named sections
- Static files split by concern: style.css, app.js, loader.js, nav.js
- Each page section is its own partial in static/sections/s-*.html
- loader.js fetches all section partials in parallel, then loads app.js and nav.js
- All original functionality, element IDs, and API endpoints preserved

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-25 19:53:01 +02:00